CN117827613A - Performance index detection method, system, equipment and storage medium - Google Patents
Performance index detection method, system, equipment and storage medium Download PDFInfo
- Publication number
- CN117827613A CN117827613A CN202410005418.5A CN202410005418A CN117827613A CN 117827613 A CN117827613 A CN 117827613A CN 202410005418 A CN202410005418 A CN 202410005418A CN 117827613 A CN117827613 A CN 117827613A
- Authority
- CN
- China
- Prior art keywords
- performance index
- determining
- historical
- index
- performance
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000001514 detection method Methods 0.000 title claims abstract description 56
- 238000004458 analytical method Methods 0.000 claims abstract description 109
- 230000002159 abnormal effect Effects 0.000 claims abstract description 56
- 238000000034 method Methods 0.000 claims abstract description 42
- 230000000737 periodic effect Effects 0.000 claims description 39
- 230000008859 change Effects 0.000 claims description 22
- 238000005457 optimization Methods 0.000 claims description 14
- 230000005856 abnormality Effects 0.000 claims description 9
- 230000006870 function Effects 0.000 description 8
- 238000004891 communication Methods 0.000 description 6
- 230000008569 process Effects 0.000 description 6
- 238000010586 diagram Methods 0.000 description 5
- 230000003044 adaptive effect Effects 0.000 description 3
- 238000012423 maintenance Methods 0.000 description 3
- 238000012549 training Methods 0.000 description 3
- 238000004364 calculation method Methods 0.000 description 2
- 230000000694 effects Effects 0.000 description 2
- 238000001914 filtration Methods 0.000 description 2
- 238000011084 recovery Methods 0.000 description 2
- 238000012935 Averaging Methods 0.000 description 1
- 238000013459 approach Methods 0.000 description 1
- 238000007405 data analysis Methods 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 238000009499 grossing Methods 0.000 description 1
- 238000011835 investigation Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 230000004044 response Effects 0.000 description 1
- 230000000630 rising effect Effects 0.000 description 1
- 238000012731 temporal analysis Methods 0.000 description 1
- 238000000700 time series analysis Methods 0.000 description 1
- 238000013024 troubleshooting Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F11/00—Error detection; Error correction; Monitoring
- G06F11/30—Monitoring
- G06F11/34—Recording or statistical evaluation of computer activity, e.g. of down time, of input/output operation ; Recording or statistical evaluation of user activity, e.g. usability assessment
- G06F11/3409—Recording or statistical evaluation of computer activity, e.g. of down time, of input/output operation ; Recording or statistical evaluation of user activity, e.g. usability assessment for performance assessment
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
- G06N3/084—Backpropagation, e.g. using gradient descent
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- General Health & Medical Sciences (AREA)
- Computing Systems (AREA)
- Computational Linguistics (AREA)
- Data Mining & Analysis (AREA)
- Evolutionary Computation (AREA)
- Biomedical Technology (AREA)
- Molecular Biology (AREA)
- Biophysics (AREA)
- Artificial Intelligence (AREA)
- Life Sciences & Earth Sciences (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Health & Medical Sciences (AREA)
- Computer Hardware Design (AREA)
- Quality & Reliability (AREA)
- Debugging And Monitoring (AREA)
Abstract
The application discloses a method, a system, a device and a storage medium for detecting performance indexes, wherein the method comprises the following steps: acquiring the current performance index of a service system, and determining an intelligent model corresponding to the performance index; calculating a dynamic threshold according to the intelligent model, wherein the dynamic threshold is determined according to historical performance indexes in a historical period; and when the performance index does not meet the dynamic threshold, performing secondary analysis on the performance index, and judging whether the performance index is an abnormal index. And the accuracy of the service system performance index detection result is improved.
Description
Technical Field
The present disclosure relates to the field of data analysis technologies, and in particular, to a method, a system, an apparatus, and a storage medium for detecting a performance index.
Background
The database stability is the basis of normal operation of each service, and whether the service system can stably operate is influenced by the performance index of the database, wherein the performance index of the database comprises the number of data connection. At present, when detecting the performance index of the database, a fixed detection threshold is mainly set, and when the performance index does not meet the fixed detection threshold, the performance index is identified as an abnormal index. And detecting the performance index of the service system through a fixed detection threshold value, so that the accuracy of the performance index detection result is reduced.
Disclosure of Invention
The embodiment of the application aims to improve the accuracy of the performance index detection result of the service system by providing the performance index detection method, the system, the equipment and the storage medium.
The embodiment of the application provides a method for detecting performance indexes, which comprises the following steps:
acquiring the current performance index of a service system, and determining an intelligent model corresponding to the performance index;
calculating a dynamic threshold according to the intelligent model, wherein the dynamic threshold is determined according to historical performance indexes in a historical period;
and when the performance index does not meet the dynamic threshold, performing secondary analysis on the performance index, and judging whether the performance index is an abnormal index.
Optionally, the determining the intelligent model corresponding to the performance index includes:
acquiring historical performance indexes;
performing periodic analysis on the historical performance index to obtain a periodic analysis result, and/or performing noise analysis on the historical performance index to obtain a noise analysis result;
and determining an intelligent model corresponding to the performance index according to the periodic analysis result and/or the noise analysis result.
Optionally, the performing periodic analysis on the historical performance index to obtain a periodic analysis result includes:
acquiring the change rate of the historical performance index of the same time period in the first historical time period;
determining the similarity of the change rates of the historical performance indexes in the same time period;
determining the periodic analysis result according to the similarity;
the step of carrying out noise analysis on the historical performance indexes to obtain noise analysis results comprises the following steps:
determining the maximum value number and the minimum value number according to the historical performance index of each time period in the first historical time period, and determining the noise analysis result according to the maximum value number and the minimum value number;
or determining a data loss rate according to the historical performance index of each time period in the first historical time period, and determining the noise analysis result according to the data loss rate.
Optionally, the step of determining the intelligent model corresponding to the performance index according to the periodic analysis result and/or the noise analysis result includes:
when the periodicity analysis result is periodicity and the noise analysis result is that the noise value is larger than a preset noise value, determining that the intelligent model is a first intelligent model;
When the periodicity analysis result is periodicity and the noise analysis result is that the noise value is smaller than or equal to the preset noise value, determining that the intelligent model is a second intelligent model;
and when the periodicity analysis result is no periodicity, determining that the intelligent model is a third intelligent model.
Optionally, the step of calculating a dynamic threshold according to the smart model includes:
determining an upper limit value and a lower limit value corresponding to each time period in a second historical time period according to the intelligent model;
determining an upper bound average value according to the upper bound value corresponding to each time period, and determining a lower bound average value according to the lower bound value corresponding to each time period;
and determining the upper bound average value and the lower bound average value as dynamic thresholds corresponding to the intelligent model.
Optionally, the performing the second analysis on the performance index, determining whether the performance index is an abnormal index includes:
acquiring a historical performance index corresponding to each time period in the second historical time period;
determining a performance index mean value according to the historical performance index corresponding to each time period;
determining a first tolerance coefficient and a second tolerance coefficient according to the upper-bound average value, the performance index average value and the lower-bound average value;
Determining a first tolerance threshold according to the upper bound average value and the first tolerance coefficient, and determining a second tolerance threshold according to the lower bound average value and the second tolerance coefficient;
determining a tolerance interval and an abnormal interval according to the first tolerance threshold and the second tolerance threshold;
detecting whether the performance index is positioned in the tolerance zone or the abnormal zone;
when the performance index is located in the tolerance interval, determining that the performance index is not an abnormal index;
and when the performance index is positioned in the abnormal section, determining that the performance index is an abnormal index.
Optionally, after the step of performing the secondary analysis on the performance index and determining whether the performance index is an abnormal index, the method further includes:
when the performance index is judged to be an abnormal index, determining an optimization scheme of the abnormal index;
and optimizing the abnormality index by adopting the optimization scheme.
In addition, in order to achieve the above object, the present application further provides a performance index detection system, including:
the intelligent model determining module is used for acquiring the current performance index of the service system and determining an intelligent model corresponding to the performance index;
The dynamic threshold determining module is used for calculating a dynamic threshold according to the intelligent model, wherein the dynamic threshold is determined according to a historical performance index in a historical period;
and the abnormal index detection module is used for carrying out secondary analysis on the performance index when the performance index does not meet the dynamic threshold value, and judging whether the performance index is an abnormal index or not.
In addition, in order to achieve the above object, the present application further provides a performance index detection device, including: the method comprises the steps of a memory, a processor and a performance index detection program which is stored in the memory and can run on the processor, wherein the performance index detection program is executed by the processor to realize the performance index detection method.
In addition, in order to achieve the above object, the present application further provides a computer-readable storage medium having stored thereon a performance index detection program which, when executed by a processor, implements the steps of the performance index detection method described above.
According to the technical scheme of the performance index detection method, the system, the equipment and the storage medium, the current performance index of the service system is obtained, and the intelligent model corresponding to the performance index is determined; calculating to obtain a dynamic threshold value through the intelligent model, and determining whether the current performance index meets the dynamic threshold value; when the performance index is not met, performing secondary analysis on the performance index, and judging whether the performance index is an abnormal index or not; the dynamic threshold value is obtained by calculating the historical performance index in the historical period through the intelligent model adaptive to the current performance index, so that the threshold value is not a fixed detection threshold value any more. And meanwhile, when the performance index is detected to not meet the dynamic threshold, performing secondary analysis on the performance index so as to avoid the risk of misjudgment of the performance index caused by primary detection and improve the accuracy of the detection result of the performance index.
Drawings
FIG. 1 is a flowchart of a first embodiment of a method for detecting performance indicators according to the present application;
FIG. 2 is a flowchart of a second embodiment of a method for detecting performance indicators according to the present application;
FIG. 3 is a flowchart of a third embodiment of a method for detecting performance indicators according to the present disclosure;
FIG. 4 is a functional block diagram of a system for detecting performance indicators according to the present application;
fig. 5 is a schematic structural diagram of a hardware running environment according to an embodiment of the present application.
The achievement of the objects, functional features and advantages of the present application will be further described with reference to embodiments, with reference to the accompanying drawings, which are only illustrations of one embodiment, and not all of the applications.
Detailed Description
In order to better understand the above technical solution, exemplary embodiments of the present disclosure will be described in more detail below with reference to the accompanying drawings. While exemplary embodiments of the present disclosure are shown in the drawings, it should be understood that the present disclosure may be embodied in various forms and should not be limited to the embodiments set forth herein. Rather, these embodiments are provided so that this disclosure will be thorough and complete, and will fully convey the scope of the disclosure to those skilled in the art.
Aiming at the problem of low accuracy of the performance index detection result of the service system, the application provides a performance index detection method, which mainly comprises the following steps: acquiring a current performance index of a service system, and determining an intelligent model corresponding to the performance index, wherein the performance index comprises a database performance index and a middleware performance index; calculating a dynamic threshold according to the intelligent model, wherein the dynamic threshold is determined according to historical performance indexes in a historical period; and when the performance index does not meet the dynamic threshold, performing secondary analysis on the performance index, and judging whether the performance index is an abnormal index. The dynamic threshold value is obtained by calculating the historical performance index in the historical period through the intelligent model adaptive to the current performance index, so that the threshold value is not a fixed detection threshold value any more. And meanwhile, when the performance index is detected to not meet the dynamic threshold, performing secondary analysis on the performance index so as to avoid the risk of misjudgment of the performance index caused by primary detection and improve the accuracy of the detection result of the performance index.
In addition, in the related art, when an abnormality occurs in database connection, the troubleshooting takes a long time. Logging in different applications, checking configuration irrational and reporting errors of log files on a host deployed by the middleware, summarizing inference analysis, and uncontrollable service recovery time. The method combines periodicity of performance data with noise analysis and various intelligent operation and maintenance AI algorithms, automatically and intelligently discovers the abnormality of the database connection number in real time, and improves the detection efficiency of abnormal indexes compared with manual abnormal index investigation.
In addition, the method and the device also read performance indexes related to the middleware resources of the database, analyze periodicity and noise of historical performance indexes within 7 or more continuous days in an accumulated mode, automatically adapt to an AI algorithm according to analysis results, and generate dynamic thresholds through a training model. And then matching and detecting the index data accessed in real time with the dynamic threshold value, and carrying out secondary analysis on the index exceeding the dynamic threshold value. After the secondary analysis, determining that the abnormality does exist, diagnosing the abnormality, if the abnormality exists, the operation of the service system is not influenced, and the operation can be temporarily not processed, otherwise, diagnosing the abnormality type. And then automatically matching the optimization scheme and automatically carrying out the optimization operation, and sending an execution result to operation and maintenance personnel for checking, so as to determine whether the corresponding operation completes the recovery of the index abnormal condition, thereby avoiding the occurrence of faults or reducing the influence of the faults and ensuring the normal operation of a service system.
As shown in fig. 1, in the first embodiment of the present application, the method for detecting a performance index of the present application may be applied to a performance index detection device, where the performance index detection device may be a device such as a smart phone, a smart computer, or other terminal devices with a performance index detection function. Specifically, the method for detecting the performance index of the present application includes the following steps:
step S110, the current performance index of the service system is obtained, and an intelligent model corresponding to the performance index is determined.
In this embodiment, the current performance index of the service system includes a database performance index and a middleware performance index. The database performance index includes, but is not limited to, index data of characteristics such as database connection number, connection time consumption, waiting event data, CPU utilization rate, memory utilization rate and the like, which can feed back system service, application performance and the like. The database connection number refers to the number of data tables required for connecting a database on a database system and the number of IO activities connected to the database, and is a performance bottleneck in the database system, and when the connection number is insufficient, the performance of the database is seriously affected. The middleware performance index includes, but is not limited to, index data such as middleware connection pool utilization, long connection number, short connection number, response time, etc. The current performance index of the service system can be obtained in real time, or the current performance index of the service system can be obtained according to the requirement.
In this embodiment, after the current performance index of the service system is obtained, the present application may perform periodic analysis and noise analysis on the current obtained performance index to obtain an analysis result; and automatically matching a corresponding intelligent model according to the analysis result, wherein the intelligent model is used for calculating to obtain a dynamic threshold value. The intelligent models corresponding to different analysis results are different, so that the optimal intelligent model can be matched according to the characteristics of different performance indexes, and the accuracy of a dynamic threshold value determined later is improved.
In this embodiment, the smart models include, but are not limited to, a CVAE smart model, a KDE smart model, and a MA smart model, where each smart model has various advantages. When the performance index is periodic and the noise is large, the CVAE intelligent model is applicable. And when the performance index is periodic and the noise is smaller, the KDE intelligent model is applicable. And when the performance index is not periodic, the MA intelligent model is applicable. The intelligent model is generated according to historical performance index training. Each smart model will be described in detail below:
first: CVAE intelligent model. CVAE (Conditional Variational Autoencoder) is an extension of the variational self-encoder (Variational Autoencoder, VAE) which can generate samples with specific properties by given conditions. The CVAE adds conditions to the VAE, enabling data generation according to given conditions. For example, the properties or labels of the samples may be entered as conditions to generate samples with specified properties or labels. CVAE adds a condition input in both the encoder and decoder of the VAE, which allows it to model conditions in potential space and generate samples with specific properties.
The training process of the CVAE comprises the following steps:
(1) The condition and performance index data are taken as inputs to the network, and the input performance index data are mapped to potential representations in potential space by the encoder.
(2) A random vector is sampled in the potential space and mapped back to samples in the data space of the performance index by the decoder.
(3) And calculating reconstruction loss and KL divergence for optimizing model parameters.
(4) Model parameters are updated using back propagation.
Second,: KDE intelligent model. KDE (Kernel Density Estimation) smart model is a non-parametric approach for probability density estimation. It estimates its probability density distribution by kernel-function weighting the observed performance indicators. In the KDE model, each observed data point is considered a kernel center and the shape and bandwidth of the kernel is determined by calculating the distance between each data point and the other data points. Common kernel functions include gaussian kernel functions, epanechnikov kernel functions, and the like. The bandwidth parameter controls the width of the kernel function, and influences the weight and smoothness of the kernel function on the sample points. The core idea of the KDE model is to weight and sum the kernel functions of each observed data point to form an overall probability density estimate. By using more kernel functions, smoother probability density estimates can be obtained. Wherein each observation data point characterizes one performance indicator data.
Third,: MA intelligent model. The MA smart model is generally referred to as a Moving Average (Moving Average) model. Moving averages are a model commonly used in time series analysis for smoothing and predicting performance indicators. The basic idea of the MA model is to decompose the time series into three parts, trend, season and residual, which can be obtained by removing the observations from the trend and season. The residual sequence is then smoothed using a moving average and predicted on the basis of this. Specifically, the MA model is calculated as follows:
(1) Calculating a residual sequence: and decomposing the original time sequence into three parts of trend, season and residual error to obtain a residual error sequence.
(2) Calculating a moving average: and carrying out moving average calculation on the residual sequence, wherein the window size of the average value can be selected according to actual conditions.
(3) And (3) predicting: a moving average model is used to predict performance metrics for future points in time.
And step S120, calculating a dynamic threshold according to the intelligent model, wherein the dynamic threshold is determined according to the historical performance index in the historical period.
In this embodiment, after determining the intelligent model corresponding to the current performance index, the historical performance index of the current performance index is calculated through the intelligent model to obtain the dynamic threshold. There is a corresponding dynamic threshold for each performance indicator. The history period may be a period in which the current time passes for 24 hours. The historical performance index in the past 24 hours is obtained, and the dynamic threshold is calculated according to the historical performance index in the historical period, so that the threshold is no longer a fixed detection threshold, and the accuracy of the performance index detection result is improved.
In this embodiment, the model generation may be performed by accessing the accumulated historical index data in batch through the streaming process, and then the dynamic threshold calculation of each index may be performed by the intelligent models MA, KDE and CVAE.
And step S130, when the performance index does not meet the dynamic threshold, performing secondary analysis on the performance index to judge whether the performance index is an abnormal index.
In this embodiment, the performance index is compared with the dynamic threshold value, so as to determine whether the performance index satisfies the dynamic threshold value. And if the current performance index meets the dynamic threshold, indicating that the current performance index is a normal performance index. If the current performance index does not meet the dynamic threshold, it indicates that the current performance index may be an abnormal performance index, and secondary analysis is required for the performance index that does not meet the dynamic threshold.
When a plurality of performance indexes exist, each performance index is compared with a corresponding dynamic threshold value respectively, so that whether each performance index meets the corresponding dynamic threshold value is determined. And if the current performance index meets the dynamic threshold, indicating that the current performance index is a normal performance index. If the current performance index does not meet the dynamic threshold, the current performance index is possibly an abnormal performance index, and secondary analysis is needed to be carried out on the performance index which does not meet the dynamic threshold, so that the risk of false detection of the performance index is reduced.
In this embodiment, the secondary analysis of the performance index which does not meet the dynamic threshold may be performed by using a box-line diagram filtering method, and the manner of performing the secondary analysis by using the box-line diagram filtering method may refer to the subsequent embodiment, which will not be described herein, so as to reduce the risk of false detection of the performance index by performing the secondary analysis of the performance index which does not meet the dynamic threshold.
According to the technical scheme, the embodiment obtains the current performance index of the service system and determines the intelligent model corresponding to the performance index; calculating a dynamic threshold value through the intelligent model, and determining performance indexes which do not meet the dynamic threshold value; finally, performing secondary analysis on the performance index which does not meet the dynamic threshold value, and judging whether the performance index is an abnormal index or not; the dynamic threshold value is obtained by calculating the historical performance index in the historical period through the intelligent model adaptive to the current performance index, so that the threshold value is not a fixed detection threshold value any more. And meanwhile, when the performance index is detected to not meet the dynamic threshold, performing secondary analysis on the performance index so as to avoid the risk of misjudgment of the performance index caused by primary detection and improve the accuracy of the detection result of the performance index.
Further, referring to fig. 2, in a second embodiment of the present application, determining the smart model corresponding to the performance index in step S110 includes the following steps:
step S111, obtaining historical performance indexes.
In this embodiment, the intelligent model is trained according to the historical performance index, so that the historical performance index is obtained here to generate the intelligent model corresponding to the performance index.
Step S112, periodically analyzing the historical performance index to obtain a periodic analysis result, and/or performing noise analysis on the historical performance index to obtain a noise analysis result.
In this embodiment, the periodic analysis result includes that the historical performance index has periodicity or that the historical performance index has no periodicity. The noise analysis result includes that the noise value of the historical performance index is greater than a preset noise value, or the noise value of the historical performance index is less than or equal to the preset noise value. The preset noise value can be set according to actual conditions.
And step S113, determining an intelligent model corresponding to the performance index according to the periodic analysis result and/or the noise analysis result.
In this embodiment, the smart models include, but are not limited to, a CVAE smart model, a KDE smart model, and a MA smart model, where each smart model has various advantages. When the performance index is periodic and the noise is large, the CVAE intelligent model is applicable. And when the performance index is periodic and the noise is smaller, the KDE intelligent model is applicable. And when the performance index is not periodic, the MA intelligent model is applicable. After the periodic analysis and noise analysis are carried out on the historical performance indexes, a proper intelligent model is matched, and the accuracy of the abnormal performance index detection result is improved.
Optionally, performing periodic analysis on the historical performance index to obtain a periodic analysis result includes:
step S1121, a change rate of the historical performance index of the same period in the first historical period is acquired.
In the present embodiment, the first history period may be 7 days or more. The same time period in the first historical period is the same time period of each day, namely 11 points to 12 points of each day. The method and the device have the advantages that data change judgment with higher ring ratio similarity in the same time period in the first historical time period is periodic, and data change judgment with reduced ring ratio similarity is not periodic.
In this embodiment, acquiring the rate of change of the historical performance index for the same period of time in the first historical period includes: and determining the average value corresponding to the historical time period according to the historical performance index corresponding to each time period in a certain historical time period. And comparing each time period in the historical time period with the average value to obtain the change rate of each time period.
Wherein, the historical period may be a certain day in the first historical period, and each hour in the day is regarded as each time period in the certain historical period, and each time period has a corresponding historical performance index. The method comprises the steps of collecting historical performance indexes of each time period, calculating a mean value, and comparing the historical performance indexes of each time period in a certain historical time period with the mean value to obtain the change rate of the historical performance indexes of each time period in the certain historical time period. Wherein the change rate has a positive or negative score, and when the change rate of the historical performance index is smaller than zero, the change rate indicates a falling change, and when the change rate of the historical performance index is larger than zero, the change rate indicates a rising change.
In step S1122, the similarity of the change rates of the history performance indexes in the same period is determined.
Step S1123, determining the periodic analysis result according to the similarity.
In this embodiment, the similarity is compared with a predetermined similarity, and when the similarity is greater than or equal to the predetermined similarity, it indicates that the periodicity analysis result is periodic. And when the similarity is smaller than the preset similarity, the periodicity analysis result is no periodicity. The preset similarity may be set according to practical situations, for example, set to 20%.
For example, the sheets are calculated separatelyAverage value of historical performance indexes in dayThen all the historical performance indexes and the average value of each time period in the current day are combined>Comparing, calculating the change rate of each time period +.>(/><0 is a decrease change, < >>>0 is the rise change). And finally, comparing the similarity of the change rates of the time points of 7 days or more, and if the similarity is greater than a preset similarity, for example, greater than or equal to 20%, determining that the cycle is present. If less than 20%, no periodicity is considered.
According to the technical scheme, the method and the device for determining the characteristics of the historical performance analysis data through the periodic analysis of the historical performance indexes improve the accuracy of the intelligent model determined subsequently.
Optionally, performing noise analysis on the historical performance index to obtain a noise analysis result includes:
step S1124, determining the maximum number and the minimum number according to the historical performance index of each time period in the first historical time period, and determining the noise analysis result according to the maximum number and the minimum number.
In this embodiment, an average value is calculated according to the historical performance index of each time period in the first historical period, and the historical performance index data of a first preset proportion greater than or equal to the average value in each time period in the first historical period is determined as a maximum value. And determining the historical performance index data of a first preset proportion smaller than the average value in each time period in the first historical time period as a minimum value. Wherein the first preset ratio may be set to 80%.
In this embodiment, the number of extremums is determined according to the number of maxima and the number of minima, and when the number of extremums is greater than a second preset proportion of the total performance index data, it is determined that the noise analysis result is that the noise is greater than the preset noise value. And when the number of the extreme values is smaller than a second preset proportion of the total performance index data, determining that the noise analysis result is that the noise is smaller than or equal to a preset noise value. Wherein the second preset proportion may be set to 10%.
Or, in step S1125, a data loss rate is determined according to the historical performance index of each time period in the first historical time period, and the noise analysis result is determined according to the data loss rate.
In this embodiment, assuming that the first history period is 7 days, 1 minute and 1 data point are taken, and a total of 10080 data points are taken for 7 days, the data loss rate can be determined according to the ratio of the loss points to the total points. When the data loss rate is larger than a third preset proportion of the total points, determining that the noise analysis result is that the noise is larger than a preset noise value; and when the data loss rate is smaller than or equal to a third preset proportion of the total points, determining that the noise analysis result is that the noise is smaller than or equal to a preset noise value. Wherein the third preset proportion may be set to 20%.
Optionally, step S113 includes the steps of:
step S1131, when the periodic analysis result is periodic and the noise analysis result is that the noise value is greater than the preset noise value, determining that the intelligent model is a first intelligent model;
step S1132, when the periodic analysis result is periodic and the noise analysis result is that the noise value is less than or equal to the preset noise value, determining that the intelligent model is a second intelligent model;
And step S1133, when the periodicity analysis result is no periodicity, determining that the intelligent model is a third intelligent model.
In this embodiment, the preset noise value may be set according to the actual situation. Smart models include, but are not limited to, CVAE smart models, KDE smart models, and MA smart models, where each smart model has various advantages. When the performance index is periodic and the noise is large, the CVAE intelligent model is applicable. And when the performance index is periodic and the noise is smaller, the KDE intelligent model is applicable. And when the performance index is not periodic, the MA intelligent model is applicable. The first intelligent model is a CVAE intelligent model; the second intelligent model is a KDE intelligent model; the third smart model is a MA smart model.
According to the technical scheme, the intelligent model can be determined according to the periodic analysis result and the noise analysis result, and a proper intelligent model can be matched according to the characteristics of the historical performance data, so that the accuracy of the determined intelligent model is improved.
Further, referring to fig. 3, based on any of the above embodiments, in a third embodiment of the present application, step S120 includes the steps of:
step S121, determining an upper bound value and a lower bound value corresponding to each time period in the second history time period according to the intelligent model.
In this embodiment, the historical performance index corresponding to each time period in the second historical time period may be a historical performance index corresponding to each hour in the last 24 hours of the current time. There may again be multiple points in time per hour, so the upper and lower bounds in each hour may be obtained. The upper limit value refers to the maximum index value in the historical performance index data collected at each time point in the hour, and the lower limit value refers to the minimum index value in the historical performance index data collected at each time point in the hour. For example, assuming the performance index is the number of database connections, the upper bound may be the maximum number of database connections collected during the hour and the lower bound may be the minimum number of database connections collected during the hour.
Step S122, an upper bound average value is determined according to the upper bound value corresponding to each time period, and a lower bound average value is determined according to the lower bound value corresponding to each time period.
In this embodiment, an average value is obtained for the upper bound value corresponding to each time period, so as to obtain an upper bound average value. And calculating an average value of the lower bound value corresponding to each time period to obtain a lower bound average value. And the accuracy of the determined dynamic threshold value is improved by solving an upper bound average value and a lower bound average value.
And step 123, determining the upper-bound average value and the lower-bound average value as dynamic thresholds corresponding to the intelligent model.
In this embodiment, the dynamic threshold is calculated after the historical performance index is input into the corresponding intelligent model. And if the current performance index is greater than or equal to the lower-bound average value and less than or equal to the upper-bound average value, indicating that the performance index is a normal performance index. If the current performance index is smaller than the lower average value or the current performance index is larger than the upper average value, the current performance index is possibly an abnormal performance index.
Optionally, the interval surrounded by the upper-bound average value and the lower-bound average value may be regarded as a normal performance index interval, and if the current performance index is located in the normal performance index interval, the current performance index is indicated to be a normal performance index. Otherwise, if the current performance index is not in the normal performance index interval (i.e. does not meet the dynamic threshold), it indicates that the current performance index is not the normal performance index.
According to the technical scheme, the upper bound average value is obtained by averaging the upper bound value corresponding to each time period. And calculating an average value of the lower bound value corresponding to each time period to obtain a lower bound average value. And the accuracy of the determined dynamic threshold value is improved by solving an upper bound average value and a lower bound average value.
Further, based on the third embodiment, in a fourth embodiment of the present application, step S130 includes the steps of:
step S131, obtaining a historical performance index corresponding to each time period in the second historical time period.
Step S132, determining a performance index mean value according to the historical performance index corresponding to each time period.
In this embodiment, the historical performance index corresponding to each time period in the second historical time period may be a historical performance index corresponding to each hour in the last 24 hours of the current time.
Step S133, determining a first tolerance coefficient and a second tolerance coefficient according to the upper-bound average value, the performance index average value and the lower-bound average value.
In this embodiment, if the current performance index does not meet the dynamic threshold, a second analysis is required to be performed on the performance index, and before the second analysis, a tolerance interval and an abnormal interval are required to be determined, and which interval the current performance index is located in is determined, so as to determine whether the current performance index is abnormal, and improve accuracy of performance index detection results.
In this embodiment, a difference between the upper bound mean and the performance index mean may be determined, and the first tolerance coefficient may be determined according to a ratio between the difference and the performance index mean. And determining a difference value between the performance index mean value and the lower bound mean value, and determining a second tolerance coefficient according to a ratio of the difference value to the performance index mean value. The first tolerance coefficient and the second tolerance coefficient are updated along with the updating of the historical performance index, so that the determined coefficient is dynamically changed.
Step S134, determining a first tolerance threshold according to the upper-bound average and the first tolerance coefficient, and determining a second tolerance threshold according to the lower-bound average and the second tolerance coefficient.
In this embodiment, the product between the upper bound mean and the first tolerance coefficient may be determined as the first tolerance threshold. The product between the lower bound mean and the second tolerance coefficient is determined as a second tolerance threshold. Wherein, because the upper bound average is greater than the lower bound average, the first tolerance threshold is greater than the second tolerance threshold.
And step S135, determining a tolerance interval and an abnormal interval according to the first tolerance threshold and the second tolerance threshold.
In this embodiment, a section defined by the first tolerance threshold and the upper-bound mean value, and a section defined by the second tolerance threshold and the lower-bound mean value are determined as tolerance sections. And determining a section which is larger than the first tolerance threshold or smaller than the second tolerance threshold as an abnormal section. The first tolerance threshold is larger than the upper-bound average value, and the second tolerance threshold is smaller than the lower-bound average value. According to the method and the device, the tolerance interval and the abnormal interval are determined, so that the accuracy of the secondary analysis result of the performance index can be improved.
Step S136, detecting whether the performance index is located in the tolerance zone or the abnormal zone.
And step S137, when the performance index is located in the tolerance interval, determining that the performance index is not an abnormal index.
And step S138, determining that the performance index is an abnormal index when the performance index is located in the abnormal section.
In this embodiment, assuming that the upper-bound mean is AIUpper and the lower-bound mean is AIlower, the normal interval is. If the current performance index is->If the performance index is normal, no secondary analysis is needed, otherwise, if the current performance index does not belong to +.>The performance index may be an anomaly index and require a secondary analysis.
Assuming that the first tolerance coefficient is T1, the second tolerance coefficient is T2, wherein,,/>. Then the first tolerance threshold is +.>The second tolerance threshold is->. Tolerance interval is +.>Or->. If the current performance index is located +.>Or alternativelyAnd when the current performance index is the normal performance index.
The abnormal section isAnd->If the current performance index is located atOr->The current performance index is indicated as an abnormal performance index.
According to the technical scheme, the current performance index is detected by setting the tolerance interval and the normal interval, so that false detection of the performance index is avoided.
Further, based on any of the above embodiments, in a fifth embodiment of the present application, after step S130, the following steps are further included:
step S210, when the performance index is judged to be an abnormal index, determining an optimization scheme of the abnormal index.
And step S220, optimizing the abnormality index by adopting the optimization scheme.
In this embodiment, matching of the optimization scheme is automatically performed according to the result type of the secondary analysis, and the optimization operation is automatically completed after the matching is completed, so that the performance index value is restored to be within the normal range. And finally updating the output information after execution to an optimization report to complete the optimization flow. Meanwhile, each index value can be self-defined and adjusted according to expert experience. The result type data table is as follows:
optionally, checking and confirming whether the optimization scheme is successfully executed, whether the abnormal event is eliminated, and whether the corresponding related index is restored to the normal level.
Optionally, automatically sending a mail and a short message to inform operation and maintenance personnel of the fault self-healing event.
According to the technical scheme, the method and the device can determine the optimization scheme of the abnormal index, optimize the abnormal index and enable the service system to resume operation as soon as possible.
The embodiments of the present application provide embodiments of a method for detecting a performance index, and it should be noted that, although a logic sequence is shown in the flowchart, in some cases, the steps shown or described may be performed in a different sequence than that shown or described herein.
As shown in fig. 4, the performance index detection system provided in the present application includes:
the intelligent model determining module 10 is used for acquiring the current performance index of the service system and determining an intelligent model corresponding to the performance index;
a dynamic threshold determining module 20, configured to calculate a dynamic threshold according to the intelligent model, where the dynamic threshold is determined according to a historical performance index in a historical period;
and the abnormal index detection module 30 is configured to perform secondary analysis on the performance index when the performance index does not meet the dynamic threshold, and determine whether the performance index is an abnormal index.
The specific implementation manner of the performance index detection system is basically the same as that of each embodiment of the performance index detection method, and is not repeated here.
As shown in fig. 5, fig. 5 is a schematic structural diagram of a hardware operating environment of a performance index detection device according to an embodiment of the present application. The performance index detection apparatus may include: a processor 1001, such as a CPU, memory 1005, user interface 1003, network interface 1004, communication bus 1002. Wherein the communication bus 1002 is used to enable connected communication between these components. The user interface 1003 may include a display, an input unit such as a keyboard, and the optional user interface 1003 may also include a standard wired interface, a wireless interface. The network interface 1004 may optionally include a standard wired interface, a wireless interface (e.g., WI-FI interface). The memory 1005 may be a high-speed RAM memory or a stable memory such as a disk memory. The memory 1005 may also optionally be a storage device separate from the processor 1001 described above.
It will be appreciated by those skilled in the art that the configuration of the performance level detection apparatus shown in fig. 5 is not limiting of the performance level detection apparatus and may include more or fewer components than shown, or may be a combination of certain components, or may be a different arrangement of components.
As shown in fig. 5, the memory 1005, which is one type of storage medium, may include an operating system, a network communication module, a user interface module, and a performance index detection program. The operating system is a program for managing and controlling the hardware and software resources of the performance index detection device, the performance index detection program and other software or program operations.
In the performance index detection apparatus shown in fig. 5, the user interface 1003 is mainly used for connecting a terminal, and performs data communication with the terminal; the network interface 1004 is mainly used for a background server and is in data communication with the background server; the processor 1001 may be used to invoke a detection program of the performance indicators stored in the memory 1005.
In this embodiment, the performance index detection apparatus includes: a memory 1005, a processor 1001, and a performance index detection program stored on the memory and executable on the processor, wherein:
When the processor 1001 calls a detection program of the performance index stored in the memory 1005, the following operations are performed:
acquiring the current performance index of a service system, and determining an intelligent model corresponding to the performance index;
calculating a dynamic threshold according to the intelligent model, wherein the dynamic threshold is determined according to historical performance indexes in a historical period;
and when the performance index does not meet the dynamic threshold, performing secondary analysis on the performance index, and judging whether the performance index is an abnormal index.
Based on the same inventive concept, the embodiments of the present application further provide a computer readable storage medium, where the computer readable storage medium stores a performance index detection program, where each step of the performance index detection method described above is implemented when the performance index detection program is executed by a processor, and the same technical effects can be achieved, so that repetition is avoided, and no redundant description is provided herein.
Because the storage medium provided in the embodiments of the present application is a storage medium used for implementing the method in the embodiments of the present application, based on the method described in the embodiments of the present application, a person skilled in the art can understand the specific structure and the modification of the storage medium, and therefore, the description thereof is omitted herein. All storage media used in the methods of the embodiments of the present application are within the scope of protection intended in the present application.
It should be noted that, in this document, the terms "comprises," "comprising," or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or system that comprises a list of elements does not include only those elements but may include other elements not expressly listed or inherent to such process, method, article, or system. Without further limitation, an element defined by the phrase "comprising one … …" does not exclude the presence of other like elements in a process, method, article, or system that comprises the element.
The foregoing embodiment numbers of the present application are merely for describing, and do not represent advantages or disadvantages of the embodiments.
From the above description of the embodiments, it will be clear to those skilled in the art that the above-described embodiment method may be implemented by means of software plus a necessary general hardware platform, but of course may also be implemented by means of hardware, but in many cases the former is a preferred embodiment. Based on such understanding, the technical solution of the present application may be embodied essentially or in a part contributing to the prior art in the form of a software product stored in a storage medium (such as ROM/RAM, magnetic disk, optical disk) as described above, including several instructions for causing a terminal device (which may be a mobile phone, a computer, a server, a television, or a network device, etc.) to perform the method described in the embodiments of the present application.
The foregoing description is only of the preferred embodiments of the present application, and is not intended to limit the scope of the claims, and all equivalent structures or equivalent processes using the descriptions and drawings of the present application, or direct or indirect application in other related technical fields are included in the scope of the claims of the present application.
Claims (10)
1. The method for detecting the performance index is characterized by comprising the following steps of:
acquiring the current performance index of a service system, and determining an intelligent model corresponding to the performance index;
calculating a dynamic threshold according to the intelligent model, wherein the dynamic threshold is determined according to historical performance indexes in a historical period;
and when the performance index does not meet the dynamic threshold, performing secondary analysis on the performance index, and judging whether the performance index is an abnormal index.
2. The method for detecting a performance index according to claim 1, wherein determining the intelligent model corresponding to the performance index comprises:
acquiring historical performance indexes;
performing periodic analysis on the historical performance index to obtain a periodic analysis result, and/or performing noise analysis on the historical performance index to obtain a noise analysis result;
And determining an intelligent model corresponding to the performance index according to the periodic analysis result and/or the noise analysis result.
3. The method for detecting a performance index according to claim 2, wherein the periodically analyzing the historical performance index to obtain a periodic analysis result includes:
acquiring the change rate of the historical performance index of the same time period in the first historical time period;
determining the similarity of the change rates of the historical performance indexes in the same time period;
determining the periodic analysis result according to the similarity;
the step of carrying out noise analysis on the historical performance indexes to obtain noise analysis results comprises the following steps:
determining the maximum value number and the minimum value number according to the historical performance index of each time period in the first historical time period, and determining the noise analysis result according to the maximum value number and the minimum value number;
or determining a data loss rate according to the historical performance index of each time period in the first historical time period, and determining the noise analysis result according to the data loss rate.
4. The method for detecting a performance index according to claim 2, wherein the step of determining the intelligent model corresponding to the performance index according to the periodic analysis result and/or the noise analysis result includes:
When the periodicity analysis result is periodicity and the noise analysis result is that the noise value is larger than a preset noise value, determining that the intelligent model is a first intelligent model;
when the periodicity analysis result is periodicity and the noise analysis result is that the noise value is smaller than or equal to the preset noise value, determining that the intelligent model is a second intelligent model;
and when the periodicity analysis result is no periodicity, determining that the intelligent model is a third intelligent model.
5. The method of claim 1, wherein the step of calculating a dynamic threshold from the smart model comprises:
determining an upper limit value and a lower limit value corresponding to each time period in a second historical time period according to the intelligent model;
determining an upper bound average value according to the upper bound value corresponding to each time period, and determining a lower bound average value according to the lower bound value corresponding to each time period;
and determining the upper bound average value and the lower bound average value as dynamic thresholds corresponding to the intelligent model.
6. The method for detecting a performance index according to claim 5, wherein said performing a secondary analysis on the performance index to determine whether the performance index is an abnormal index comprises:
Acquiring a historical performance index corresponding to each time period in the second historical time period;
determining a performance index mean value according to the historical performance index corresponding to each time period;
determining a first tolerance coefficient and a second tolerance coefficient according to the upper-bound average value, the performance index average value and the lower-bound average value;
determining a first tolerance threshold according to the upper bound average value and the first tolerance coefficient, and determining a second tolerance threshold according to the lower bound average value and the second tolerance coefficient;
determining a tolerance interval and an abnormal interval according to the first tolerance threshold and the second tolerance threshold;
detecting whether the performance index is positioned in the tolerance zone or the abnormal zone;
when the performance index is located in the tolerance interval, determining that the performance index is not an abnormal index;
and when the performance index is positioned in the abnormal section, determining that the performance index is an abnormal index.
7. The method for detecting a performance index according to claim 1, wherein after the step of performing the secondary analysis on the performance index to determine whether the performance index is an abnormal index, further comprising:
when the performance index is judged to be an abnormal index, determining an optimization scheme of the abnormal index;
And optimizing the abnormality index by adopting the optimization scheme.
8. A performance index detection system, wherein the performance index detection system comprises:
the intelligent model determining module is used for acquiring the current performance index of the service system and determining an intelligent model corresponding to the performance index;
the dynamic threshold determining module is used for calculating a dynamic threshold according to the intelligent model, wherein the dynamic threshold is determined according to a historical performance index in a historical period;
and the abnormal index detection module is used for carrying out secondary analysis on the performance index when the performance index does not meet the dynamic threshold value, and judging whether the performance index is an abnormal index or not.
9. A performance index detection apparatus, characterized in that the performance index detection apparatus comprises: memory, a processor and a performance level detection program stored on the memory and running on the processor, which performance level detection program, when executed by the processor, implements the steps of the performance level detection method according to any one of claims 1-7.
10. A computer-readable storage medium, characterized in that a performance index detection program is stored thereon, which, when executed by a processor, implements the steps of the performance index detection method of any one of claims 1 to 7.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202410005418.5A CN117827613A (en) | 2024-01-02 | 2024-01-02 | Performance index detection method, system, equipment and storage medium |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202410005418.5A CN117827613A (en) | 2024-01-02 | 2024-01-02 | Performance index detection method, system, equipment and storage medium |
Publications (1)
Publication Number | Publication Date |
---|---|
CN117827613A true CN117827613A (en) | 2024-04-05 |
Family
ID=90507736
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202410005418.5A Pending CN117827613A (en) | 2024-01-02 | 2024-01-02 | Performance index detection method, system, equipment and storage medium |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN117827613A (en) |
-
2024
- 2024-01-02 CN CN202410005418.5A patent/CN117827613A/en active Pending
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN112162878B (en) | Database fault discovery method and device, electronic equipment and storage medium | |
CN110083507B (en) | Key performance index classification method and device | |
CN111444060B (en) | Abnormality detection model training method, abnormality detection method and related devices | |
CN116739829B (en) | Big data-based power data analysis method, system and medium | |
CN111881961A (en) | Power distribution network fault risk grade prediction method based on data mining | |
CN115794578A (en) | Data management method, device, equipment and medium for power system | |
CN108306997B (en) | Domain name resolution monitoring method and device | |
CN115878171A (en) | Middleware configuration optimization method, device, equipment and computer storage medium | |
CN114090393B (en) | Method, device and equipment for determining alarm level | |
CN114301803B (en) | Network quality detection method and device, electronic equipment and storage medium | |
CN113254250B (en) | Database server abnormal cause detection method, device, equipment and storage medium | |
CN117687884A (en) | Intelligent optimization method and system for operation and maintenance operation ticket of power grid dispatching automation master station | |
CN115114124A (en) | Host risk assessment method and device | |
CN113590427A (en) | Alarm method, device, storage medium and equipment for monitoring index abnormity | |
CN117370753A (en) | Method, system and storage medium for identifying abnormal power users based on big data | |
CN117117780A (en) | Circuit breaker anti-blocking method and system based on secondary information fusion of transformer substation | |
CN117827613A (en) | Performance index detection method, system, equipment and storage medium | |
CN114157486B (en) | Communication flow data abnormity detection method and device, electronic equipment and storage medium | |
CN111783883A (en) | Abnormal data detection method and device | |
CN115357011A (en) | Robot fault processing method, device, medium and electronic equipment | |
CN114358581A (en) | Method and device for determining abnormal threshold of performance index, equipment and storage medium | |
CN114037285A (en) | Distribution network automation application success analysis method and related system | |
CN113656452A (en) | Method and device for detecting abnormal index of call chain, electronic equipment and storage medium | |
CN114172708A (en) | Method for identifying network flow abnormity | |
CN112153685B (en) | RRC fault detection method and device |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination |